BUGSI Benchmarking Report

Date: 08 January 2025

How to use the report?

Summary

Data Matrix

Reference Match (RM): Isoform is FSM and within 50 bp of annotated transcription start and end sites.

Full Splice Match (FSM): Isoform with identical exon count and matching internal junctions; start and end may vary.

Incomplete Splice Match (ISM): Isoform missing some 5’ exons but with matching internal junctions.

Novel In Catalog (NIC): Isoform uses known splice sites but lacks FSM or ISM match.

Novel Not in Catalog (NNC): Isoform contains unannotated splice sites.

Genic Intron: Isoform entirely within an annotated intron.

Evaluation Metrics

1. Sensitivity (Sn):

\[ \text{Sn} = \frac{\text{Unique (TP)}}{\text{Total BUGSI Genes}} \]

2. Non-redundant Precision (nrPre):

\[ \text{nrPre} = \frac{\text{TP}}{\text{Total BUGSI Match Transcripts}} \]

3. Precision (Pre):

\[ \text{Pre} = \frac{\text{TP} + \text{PTP}}{\text{Total BUGSI Match Transcripts}} \]

4. Positive Detection Rate (PDR):

\[ \text{PDR} = \frac{\text{Unique (TP + PTP)}}{\text{Total BUGSI Genes}} \]

5. False Discovery Rate (FDR):

\[ \text{FDR} = \frac{\text{Total BUGSI Match Transcripts} - \text{TP}}{\text{Total BUGSI Match Transcripts}} \]

6. False Detection Rate:

\[ \text{False Detection Rate} = \frac{\text{FP}}{\text{Total BUGSI Match Transcripts}} \]

7. Redundancy (1/Red):

\[ \text{Redundancy (1/Red)} = \frac{\text{TP + PTP}}{\text{Unique (TP + PTP)}} \]

Benchmarking Result

Radar Plot

Structural Category

Decision Matrix

TP

PTP

FN

FP